Does the AI Supercycle Deliver Value or Just Burn Capital?
AI Capex: From Promise to Proof
Amazon’s $200B 2026 capex plan highlights hyperscaler spending, but combined hyperscaler investment figures remain unspecified. Alphabet’s $45B Q2 capex (100% YoY) and $195-205B FY guidance reflect aggressive investment, though first-ever negative free cash flow signals a shift toward profitability. Meta’s exploration of compute leasing (not a completed pivot) includes plans to monetize infrastructure via external clients like Anthropic. AMD’s Helios rack system, combining Venice CPUs with MI455X GPUs, targets a $2T total addressable market by 2030, with 60% of AI compute expected for inference in 2026. Nvidia’s 82% revenue growth (analyst projection) contrasts with 36% client concentration risk. Tesla’s operating margin collapse to 1.4% (from 4.1%) and $5.8B Q2 capex underscore overleveraging risks, with Optimus as a speculative driver funded by shareholder capital. Oil prices nearing $120/barrel due to Houthi attacks and geopolitical tensions create margin pressures for tech capex, benefiting upstream sectors like Exxon. GE Vernova’s $176B backlog and $5.1B Q2 FCF position it to capitalize on AI-driven grid demand. The article emphasizes capital efficiency and near-term ROI clarity over speculative bets.